Seth Redmore Chief Marketing Officer

@sredmore sredmore@lexalytics.com

Words++ : How Social Communication Methods 📊 ❤

Are Changing the Game for Market Research

Who Am I 2

With over 20 years of combined experience in product management, marketing, text analytics and machine learning, Seth is currently the CMO of text analytics leader Lexalytics. Prior to this role, Seth served as Vice President of Product Management and Vice President of Marketing at Lexalytics. Seth has also held executive positions at both hardware and software companies, including co-founder of Netiverse (acquired by Cisco Systems). During his tenure at Cisco, Seth built Cisco's first internal text analytics solution for reputation management. Seth has a degree in Chemistry from Carnegie Mellon University.

Who is Lexalytics 3

§  Founded in 2003, first product in 2004 §  Privately held §  ~150 Customers serving thousands of major brands §  “The” engine for

–  Media Monitoring –  Social Listening –  Customer Experience Management –  Voice of the Customer

§  Multiple patents in Natural Language Processing §  We process more words than anyone else.

–  Billions/day on-premise Salience –  >150 Million/day via Semantria

Applications

SaaS (Semantria)

Core NLP

(Salience)

Agenda

🕐 🌏🎌🚀🌍🌎📈 🕑 = ??? 🕒 📖 + 😃 ➡ 📖 + 😃 + 📷 + 📹 🕔 💻 👀 = ♺ 📚 🕓 🔭🔮 🔆👓

4

Emoji Usage 5

•  ~4% of all tweets have Emoji •  Kralj Novak P, Smailović J, Sluban B,

Mozetič I (2015) Sentiment of Emojis. PLoS ONE 10(12): e0144296. doi:10.1371/journal.pone.0144296

•  ~15% of all brand-related tweets have Emoji •  Lexalytics Observations

http://instagram-engineering.tumblr.com/post/117889701472/emojineering-part-1-machine-learning-for-emoji

Emoji use vs. Internet Slang Use (Instagram)

6

http://instagram-engineering.tumblr.com/post/117889701472/emojineering-part-1-machine-learning-for-emoji

7

http://grouplens.org/site-content/uploads/ICWSM16_Emoji-Final_Version.pdf

http://grouplens.org/blog/investigating-the-potential-for-miscommunication-using-emoji/

8

http://grouplens.org/blog/investigating-the-potential-for-miscommunication-using-emoji/

Common Phrases for Major Emoji

😂 (ranked 1st in emoji usage): lolol, lmao, lololol, lolz, lmfao, lmaoo, lolololol, lol, ahahah, ahahha, loll, ahaha, ahah, lmfaoo, ahha, lmaooo, lolll, lollll, ahahaha, ahhaha, lml, lmfaooo

😍 (ranked 2nd in emoji usage): beautifull, gawgeous, gorgeous, perfff, georgous, gorgous, hottt, goregous, cuteeee, beautifullll, georgeous, baeeeee, hotttt, babeee, sexyyyy, perffff, hawttt

❤ (ranked 3rd in emoji usage): xoxoxox, xoxoxo, xoxo, xoxoxoxo, xoxoxoxoxo, xoxoxoxox, xxoo, oxox, babycakes, muahhhh, mwahh, babe, boobear, loveyou, bunches, muahhh, muahh, xoxox, muahhhhh

👍(ranked 9th in emoji usage): #keepitup, #fingerscrossed, aswell, haha, #impressed, #yourock, lol, #greatjob, bud, #goodjob, awesome, good, #muchlove, #proudofyou, job, #goodluck

😭(ranked 11th in emoji usage): ughh, ughhh, ughhhh, ugh, uggh, ugghh, ughhhhh, ughhhhhh, ugggh, lolol, wahhhh, rn, oml, uhg, agh, xc, omgg, omfg, omf, lololol, whyyy, loll, wahhhhh, tooo, kms

9

http://instagram-engineering.tumblr.com/post/117889701472/emojineering-part-1-machine-learning-for-emoji

IOS 8.3 Emoji continue to mutate. 10

http://instagram-engineering.tumblr.com/post/118304328152/emojineering-part-2-implementing-hashtag-emoji

Real life Emoji – Photos and Videos.

11

Computer Vision 12

Computer Vision 13

Computer Vision/Image Recognition

§  Current state of technology is where text analytics was ~10-12 years ago. §  Point products w/point functionality

–  Very fragmented market –  Categorization –  Identification of Brands –  Facial Emotion Analysis

§  Data access questions –  No “Instagram Firehose” –  Query-based access kinda defeats the point

§  What can we learn from history? –  Do not overpromise. –  Just counting “mentions” becomes rapidly uninteresting – need other data

as well –  Verticalize sooner rather than later

•  Security vs. Media Monitoring §  One major difference: Convolutional Neural Networks

14

Convolutional Neural Networks 15

GPU 1: Trained itself (no config) on edge detection

GPU 2: Trained itself (no config) on color boundaries and textures

http://code.flickr.net/2014/10/20/introducing-flickr-park-or-bird/

http://colah.github.io/posts/2014-07-Conv-Nets-Modular/

A look at the future!

§  Emoji –  “Sentences” and interactions will be better understood –  Fonts will normalize some

§  Computer Vision –  Business use case will be sorted out –  Image Recognition

•  Classes will broaden out •  Multi-class systems will come to market

–  Video •  Will move from statistical sampling to ”better than real time”

–  Data •  Availability issues are still up in the air •  Image “search” reliant on keywords from metadata, will move

to coming from CV-based image categorization over time

16

Resources

§  http://emojipedia.org/

18